Artificial intelligence (AI) is transforming investment strategies, with tools like ChatGPT and Claude being utilized by investors to inform their decisions. Edward Morris, a technology entrepreneur, notably used ChatGPT for due diligence on the $5 billion initial public offering (IPO) of chip designer Arm, achieving a 30% return on an investment where he typically targets 10%. He credits the AI with simplifying complex wealth management topics, identifying worthy investments, and rapidly conducting due diligence that previously took days. Morris believes AI can help people make "educated investments" without the cost of professional wealth management.
Despite the apparent benefits, significant risks are associated with relying on AI for investment advice. There's a concern that AI may provide poor advice, leading to substantial financial losses, as exemplified by a 2019 incident where entrepreneur Samathur Li Kin-kan reportedly lost $20 million using a robo-investor service. These risks are amplified by "AI hallucinations"—the generation of false or fictitious information—and biases stemming from the chatbots' training data. Experts like Neil Sahota warn that AI systems often offer poor advice due to "limited personalization" and a "lack of human empathy.
Research indicates a growing reliance on AI, particularly among younger investors. A Financial Conduct Authority study found that 80% of 18- to 40-year-olds considering investments have used AI for related support, with nearly half (44%) mistakenly believing AI-generated financial information is regulated. Moreover, 38% believe it's acceptable to make investment decisions based solely on AI outputs. Experts caution that while AI can make investing more accessible by explaining complex concepts, it cannot replace regulated financial advice or personal judgment. This overconfidence, combined with low-cost trading apps and instant AI responses, could lead to dangerous outcomes.
The widespread adoption of powerful AI models like Anthropic's Claude on Wall Street could also lead to "herding behavior" and increased market volatility. As more analysts and investors use the same AI tools to process data and generate insights, they may converge on similar conclusions and investment strategies. This not only increases the risk of missing "black swan" events but also amplifies market volatility if many participants react identically to information. Federal Reserve Governor Michael Barr has warned that the ubiquitous use of generative AI tools could concentrate risk and amplify market swings, as AI models, designed to predict the most likely next word, tend to echo familiar patterns rather than generate original insights.
Initial tests of AI in investment scenarios have shown mixed results. In one trading tournament, frontier large-language models (LLMs) given $10,000 to trade US tech stocks over two weeks lost approximately one-third of the total cash, with over 80% of the portfolios ending in the red. This suggests that LLMs often "trade too much" and make inconsistent decisions, reflecting the behavior of an "average meatbag market participant." Additionally, research from Goethe University found that investors using LLMs for research were significantly more likely to buy stocks than those using conventional search engines, suggesting that chatbots may be used for "confirmation-seeking," validating pre-existing beliefs rather than providing objective advice.
